YouTube2Text

중국 2030년까지 AI데이터센터 5배 더 짓는다 한국 증시 수혜 종목 분석 — Transcript

by 소투 · 2,274 words · 410 segments · language en · Watch on YouTube

Full transcript

  1. 0:01Hello. I am a small-cap investor.
  2. 0:04Today is Sunday. On this Sunday morning
  3. 0:07, my interest
  4. 0:10is on Telegram. With 15,000 followers,
  5. 0:14the channel is doing well. Someone
  6. 0:16shared this, so I took a look, and
  7. 0:19yesterday news came out that South Korea
  8. 0:22is racing to build AI data centers.
  9. 0:25I checked the content, and it said they
  10. 0:28would increase capacity by a massive
  11. 0:30fivefold. So, I decided to make a video
  12. 0:33about it. I saw in the comments that the
  13. 0:37microphone sound wasn't very clear.
  14. 0:40So, I bought a new microphone this time
  15. 0:43. It's like a condenser mic,
  16. 0:47so I hope the sound quality improves,
  17. 0:49and I apologize for any inconvenience
  18. 0:52caused.
  19. 0:54Let's briefly go over the news and
  20. 0:58comments. Here is the news.
  21. 1:02It seems it aired on the TBS 9 o'clock
  22. 1:04news. The 9 o'clock news reports an
  23. 1:06aggressive expansion of infrastructure.
  24. 1:08It mentions "abundant electricity."
  25. 1:11This is all solar power. It's incredibly
  26. 1:14convenient since China has a lot of land
  27. 1:17. They have a lot of land, especially in
  28. 1:19the west, and the rivers.
  29. 1:21They are installing solar panels there,
  30. 1:24and the same goes for wind power in the
  31. 1:27west.
  32. 1:28Both the front and back of these solar
  33. 1:31panels require PET films.
  34. 1:34Films are used in all of them, so the
  35. 1:37current expectation is
  36. 1:42not just for GCL, a famous Korean solar
  37. 1:44company, but various solar companies.
  38. 1:49CSI, a Nasdaq-listed company with large
  39. 1:52factories, is also involved. They are
  40. 1:55also making solar panels
  41. 1:58directly.
  42. 2:02So, they say they are expanding AI
  43. 2:04infrastructure.
  44. 2:06Inside these data centers, on every wall
  45. 2:11and floor, there are films, specifically
  46. 2:15heat dissipation films, and there are
  47. 2:19chips, and high-end chips.
  48. 2:23They naturally require heat dissipation
  49. 2:25.
  50. 2:31The same applies to MLCCs.
  51. 2:36For CC film, you attach and detach it.
  52. 2:43But in the manufacturing process,
  53. 2:48currently, S Electric's costs
  54. 2:52if it passes the test, my expectations
  55. 2:57if it doesn't...
  56. 3:03They are currently testing it, and even
  57. 3:06if it's lacking, they might pass the
  58. 3:08call test.
  59. 3:10If they do, why would it be a positive
  60. 3:12factor?
  61. 3:15Because in the CCC, S Electric is the
  62. 3:18top...
  63. 3:22When going to sell MLCCs to companies
  64. 3:24like those in China or Japan, they say
  65. 3:26they check if the product is used in S
  66. 3:28Electric.
  67. 3:30It hasn't entered yet, but until...
  68. 3:35Until...
  69. 3:49Renewable energy like solar or ESS is
  70. 3:54the same.
  71. 3:57The secondary battery sector.
  72. 4:00The good thing about China is its
  73. 4:02balanced development strategy. The
  74. 4:05conclusion for the inland regions is...
  75. 4:09Looking at the US, for example, Texas,
  76. 4:12there are existing industrial and
  77. 4:15residential complexes,
  78. 4:20and they were built right next to them.
  79. 4:25First of all, the noise, the noise is
  80. 4:27severe.
  81. 4:31Electricity rates rise.
  82. 4:33Costs rise.
  83. 4:36It's not just that; it's serious, and
  84. 4:39housing prices are falling.
  85. 4:45Not only that, but invisible
  86. 4:47electromagnetic waves,
  87. 4:50waves, carbon emissions, emissions.
  88. 4:54The heat also goes up.
  89. 4:57But more importantly, since China lacks
  90. 5:00power, they say they are installing
  91. 5:03these small, makeshift gas turbines.
  92. 5:06The noise is no joke; data centers by
  93. 5:09themselves aren't very loud. But the
  94. 5:12sound of the turbines running is so loud
  95. 5:15that it's becoming a major issue.
  96. 5:21Since it's that region, there aren't
  97. 5:22many people. But it has to be built.
  98. 5:24It has to be built for power generation
  99. 5:26.
  100. 5:28In the case of the US, they should have
  101. 5:31planned it from the start to be built in
  102. 5:33deserts or inland areas.
  103. 5:36But because they built them in places
  104. 5:38like Texas where people live, it's hard
  105. 5:40to manage now, similar to these areas in
  106. 5:42Korea. Mostly here, this...
  107. 5:46Yes, it's the Inner Mongolia region.
  108. 5:47Even here...
  109. 5:50Since it's a developed region, it's
  110. 5:51impossible for there to be none at all.
  111. 5:54These are the two most developed areas
  112. 5:56around Guangzhou; this is Jiangsu
  113. 5:58Province and this is Guangdong, and
  114. 6:00their GDP is number one here.
  115. 6:03This one is first and this is second,
  116. 6:04but they are similar.
  117. 6:06They've almost caught up now, since
  118. 6:08Jiangsu is so big.
  119. 6:12In very developed regions, it's slightly
  120. 6:16lower.
  121. 6:20In a way, there's no justification to
  122. 6:22oppose it here. The citizens can't
  123. 6:24oppose it, and since it's a communist
  124. 6:25system, they wouldn't be able to anyway
  125. 6:27. It seems it will move forward at a
  126. 6:30good speed.
  127. 6:34AI data centers, that's what it is. It's
  128. 6:37right here.
  129. 6:44Since it's a positive region, there
  130. 6:45would be no reason for them to oppose
  131. 6:47this.
  132. 6:48There's no reason, no power to oppose it
  133. 6:50, and that seems to be the situation.
  134. 6:52As you can see, the AI data center
  135. 6:55capacity is set to increase fivefold,
  136. 6:58and the statement says by 2030, which is
  137. 7:02basically ten years from now, meaning
  138. 7:05they'll increase it five times over.
  139. 7:10Currently, the factories...
  140. 7:15The factories...
  141. 7:20Increasing fivefold by 2030.
  142. 7:24There are many cars, and if you look at
  143. 7:26an AI data center, you see buildings
  144. 7:29like this.
  145. 7:32Construction notes are similar, and
  146. 7:35naturally, the notes...
  147. 7:39Everything goes in there, but it's not
  148. 7:41just about what goes in.
  149. 7:45We have to deal with electromagnetic
  150. 7:46waves and block the towers.
  151. 7:48This board goes in, and the GPU goes in
  152. 7:52, and that's why we're worried about
  153. 7:55this.
  154. 7:56I was worried if five factories would
  155. 7:58generate enough revenue.
  156. 8:01Looking at it, I have the same thought.
  157. 8:05If you look closely at the loans.
  158. 8:09This is the situation in the US. In the
  159. 8:11US, the citizens are pushing back; I
  160. 8:16heard that 41 states are currently
  161. 8:20opposing it. It's a situation where
  162. 8:22there is opposition in every region.
  163. 8:25What's a shame is that it would have
  164. 8:27been better if they had built them in
  165. 8:29desert areas or places with fewer people
  166. 8:31from the start.
  167. 8:34It seems they built them in places like
  168. 8:36Texas where many people live to save on
  169. 8:39costs because there were no roads or
  170. 8:41infrastructure there, and because there
  171. 8:43were no proper regulations, they didn't
  172. 8:46even disclose information and just built
  173. 8:48them.
  174. 8:50Initially, the public didn't strongly
  175. 8:53oppose companies like Amazon coming in,
  176. 8:56for example. But after it was built, the
  177. 9:00noise was unbearable and property values
  178. 9:03plummeted.
  179. 9:06They are currently voicing their
  180. 9:06grievances.
  181. 9:09The US government and those companies
  182. 9:12...
  183. 9:15It feels like they might have rushed the
  184. 9:17administration.
  185. 9:22If they had vast land like China, they
  186. 9:24could go to unpopulated areas.
  187. 9:29Europe is the same. Even in Europe, they
  188. 9:31are...
  189. 9:33They are building them, but there are
  190. 9:35protests; Europe is facing huge backlash
  191. 9:37right now.
  192. 9:40This is a comment from that news video.
  193. 9:44"I thought there would be a lot of China
  194. 9:46here, but surprisingly, about 70%will be
  195. 9:48China."
  196. 9:49That kind of...
  197. 9:51They selected it. In that, in that...
  198. 9:58There isn't any. Regulations are, in
  199. 9:59fact...
  200. 10:02China doesn't really have regulations,
  201. 10:04so maybe that's why they plan more
  202. 10:06systematically, avoiding data center
  203. 10:08backlash.
  204. 10:10For example, in China's case, they would
  205. 10:12have places like Xinjiang.
  206. 10:15Because there's a goal for balanced
  207. 10:16development, the situation is better.
  208. 10:20They are treating it as a matter of
  209. 10:21national survival and investing in
  210. 10:22infrastructure.
  211. 10:26This isn't an excuse; I think they
  212. 10:28learned from the mistakes of the US.
  213. 10:31They made plans and built them to
  214. 10:33protect people from the impact.
  215. 10:38Unlike here, they are building in inland
  216. 10:41areas where there are fewer power plants
  217. 10:45; it's true.
  218. 10:47People in the semiconductor industry
  219. 10:50know how much demand there is;
  220. 10:52semiconductor demand is huge.
  221. 10:58None. Something like this.
  222. 11:01AI.
  223. 11:04That kind of perspective; actually, I
  224. 11:08think so too, it's inevitable.
  225. 11:11They may have defined frontier models
  226. 11:14separately, but...
  227. 11:17If you aim for 90%performance, you could
  228. 11:21call it that. Is...
  229. 11:25Now, looking at it globally, the US...
  230. 11:30Probably not to that extent, but...
  231. 11:33In the future, it might become similar
  232. 11:35or surpass it.
  233. 11:38Europe and the US have too much...
  234. 11:42Dictatorship is bound to be advantageous
  235. 11:43. In the auto industry, for instance, in
  236. 11:45autonomous driving...
  237. 11:49They are doing it, right?
  238. 11:51There are errors that can cause
  239. 11:52accidents. Korea keeps rejecting this
  240. 11:56due to accidents, which is a bit
  241. 12:00regrettable, and whether it's right to
  242. 12:05let Hyundai Motor do it or not...
  243. 12:11It's a field that needs development.
  244. 12:16Yes. So, now...
  245. 12:20I saw this kind of comment, this kind of
  246. 12:22viewpoint.
  247. 12:23Even the US Democratic Party cannot
  248. 12:27compete with the data center...While
  249. 12:31this is true, these people prioritize
  250. 12:35seizing power over the competition with
  251. 12:40China. In a democracy, they don't even
  252. 12:42think about the bigger picture. For
  253. 12:44example, with China...
  254. 12:47If they can get elected...
  255. 12:50Even if it means failing, they are the
  256. 12:52type of people who would do it if it
  257. 12:53meant seizing power. Korea is the same.
  258. 13:01Some...China...
  259. 13:05Whether they do well or the US does well
  260. 13:06, as long as we do well...
  261. 13:13Open weights, that is now...the same
  262. 13:14thing.
  263. 13:15In fact, China's open-weight models,
  264. 13:19like DeepSeek, are about ten times more
  265. 13:22cost-effective. Due to security issues,
  266. 13:26in about two or three years,
  267. 13:31TV users will use local AI instead,
  268. 13:35which is why the current status might
  269. 13:40change.
  270. 13:44That's why East Asia has such
  271. 13:45well-developed power infrastructure.
  272. 13:49Isn't it because the US doesn't have
  273. 13:51infrastructure like this?
  274. 13:52What surprised me was that I recently
  275. 13:55visited the Yisu industrial complex in
  276. 13:59China, where they carve out mountains to
  277. 14:02build these industrial zones.
  278. 14:06They are building a GRT factory at the
  279. 14:09top of a mountain, yet the power grid is
  280. 14:13already fully equipped.
  281. 14:15If you watch the video, there's a scene
  282. 14:19where I'm shocked by the power grid;
  283. 14:23it’s in a remote rural area, in an
  284. 14:26industrial park carved out of a mountain
  285. 14:30, and I was amazed at how well-developed
  286. 14:34the power grid was.
  287. 14:46I think that power grid will work out.
  288. 14:50It’s an AI-driven battle.
  289. 14:55This is content I left on Telegram.
  290. 14:57By 2030, it could grow fivefold.
  291. 15:04With five factories, sales are bound to
  292. 15:08increase. Not only that, there are
  293. 15:10semiconductors and many other things.
  294. 15:15There are robots, too.
  295. 15:18There are cars as well.
  296. 15:21With batteries and such a great
  297. 15:24portfolio, it’s hard to imagine a sharp
  298. 15:28drop in sales; they went public in 2016
  299. 15:31, so it's been about 10 years.
  300. 15:36Sales have never dropped, except for one
  301. 15:38time. When was it? It dipped briefly
  302. 15:40during COVID. But they still managed to
  303. 15:43maintain profits. They didn't record a
  304. 15:45loss. Even during the COVID period, it
  305. 15:47dipped briefly and then went back up.
  306. 15:49Even if something bigger than COVID
  307. 15:52happens, for example, a local war
  308. 15:54between the US and China over Taiwan
  309. 15:56issues, I don't think GRT's sales would
  310. 15:59be affected.
  311. 16:01Since they do a lot of things like war,
  312. 16:05wouldn't it be quite solid? That would
  313. 16:09be a local war.
  314. 16:10If a full-scale war breaks out, where
  315. 16:12the US conducts land operations in China
  316. 16:14, then obviously that would be a
  317. 16:16different story.
  318. 16:17Even if that happens, I...
  319. 16:22Data ceiling is high.
  320. 16:25Yes.
  321. 16:30The important thing is that the
  322. 16:33specialized film is the first in China
  323. 16:36to be produced under a patent, which is
  324. 16:39what I talked about regarding the new
  325. 16:42factory.
  326. 16:45In fact, employees said those factories
  327. 16:47would have very good operating profit
  328. 16:49margins.
  329. 16:50They said it would be at least 5
  330. 16:50percentage points better. They are
  331. 16:51saying this. Of course, the factory
  332. 16:53won't improve, but automation...
  333. 16:57Even the YouTube comments are now...
  334. 17:01The atmosphere has changed so much in
  335. 17:03just one year that it's surprising.
  336. 17:05When it first came to Korea, people were
  337. 17:08mocking it heavily.
  338. 17:10It is changing quickly.
  339. 17:12The price-to-performance ratio is
  340. 17:15undeniable, even if cars are different.
  341. 17:22It seems the perception of Korea is
  342. 17:25changing.
  343. 17:26The way the U.S. is blocking China is a
  344. 17:28massive issue right now, isn't it?
  345. 17:30That's right.
  346. 17:33A lot of stocks, thousands of them, have
  347. 17:34plummeted. With the crash in Mexico and
  348. 17:37elsewhere, China is now building five
  349. 17:39times as much. Since that news is out,
  350. 17:43it’s bound to be compared.
  351. 17:46The AI infrastructure in the U.S. and
  352. 17:49Korea has surged dozens of times over.
  353. 17:52Oh, so they’re blocking this now.
  354. 17:54With China building five times more,
  355. 17:56shouldn't we invest in China? Don't you
  356. 17:58think that kind of thought comes to mind
  357. 18:00?
  358. 18:02Where will the excitement and
  359. 18:04anticipation shift toward? I believe it
  360. 18:06will shift toward Chinese AI
  361. 18:07infrastructure.
  362. 18:10I think there will be no other choice
  363. 18:12but for China to be the biggest
  364. 18:13beneficiary.
  365. 18:16So, there are things related to China's
  366. 18:19AI infrastructure, and
  367. 18:22NK Electronics as well.
  368. 18:27It's a company with a broad
  369. 18:29semiconductor portfolio; it supplies to
  370. 18:32Korean firms, but it also supplies to
  371. 18:36Chinese semiconductor firms.
  372. 18:43AC and Jangjeon, and here—
  373. 18:47I noticed this electronics company
  374. 18:48supplies to everyone globally. It’s the
  375. 18:51same for them; this electronics company
  376. 18:55is also in infrastructure, and the same
  377. 18:59goes for others, as they are fully
  378. 19:02stocked.
  379. 19:04As for the rest, like Guseong
  380. 19:06Engineering, Korean parts and equipment
  381. 19:09suppliers also supply to companies like
  382. 19:13CXMT.
  383. 19:15They could be leading stocks, but
  384. 19:17they’ve surged so much that I'm cautious
  385. 19:21.
  386. 19:22For GRT, the valuation hasn't spiked, so
  387. 19:26as for the top AI infrastructure stock,
  388. 19:30naturally—
  389. 19:33Their exposure is almost entirely to
  390. 19:34China.
  391. 19:36If so, the growth potential must be high
  392. 19:38. Their revenue and operating profit are
  393. 19:41overwhelming.
  394. 19:43With all their products in data centers
  395. 19:45, couldn't they become the leading stock
  396. 19:48?
  397. 19:50If money flows in little by little and
  398. 19:53the stock price rises, we call it the
  399. 19:55principle of trend following.
  400. 19:58With breakout trading, that kind of
  401. 20:01capital comes in, so it might
  402. 20:04consolidate, pull back, and then rise.
  403. 20:09That is what I am hoping for.
  404. 20:14I made this video in a bit of a rush
  405. 20:16over the weekend.
  406. 20:19Moving forward, I'll change my
  407. 20:20microphone and improve the video quality
  408. 20:22.
  409. 20:24Have a good time. Please leave a like
  410. 20:26and a comment.

About this transcript

This page contains the full transcript of 중국 2030년까지 AI데이터센터 5배 더 짓는다 한국 증시 수혜 종목 분석 by 소투, generated from the public captions YouTube serves with the video. The transcript has 2,274 words across 410 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.